Frevana Solution Page

Create single-file HTML landing pages that match Frevana's solution-page visual system. The user can describe the page in natural language; do not require them to hand-write JSON.

The visual target is the current Frevana /individual page: neutral white canvas, oversized Host Grotesk headlines, Open Sans body/UI text, soft-green proof chips, dark charcoal primary CTAs, green inline feature links, alternating product-illustration rows, airy spot illustrations that visually blend into the white page, and a dark slate footer.

Workflow

1. Read references/design.md before generating the page.
2. Extract the user's natural-language brief into the internal page schema.
3. Ask follow-up questions only when a missing detail would materially change the page, such as the audience, product, or required CTA URL.
4. If the user gives only CTA text, use /zh-CN/signup as the CTA URL.
5. Generate the page with scripts/generate_landing_page.py.
6. By default, generate and cache page images through the sibling gpt-image-2 skill.

Internal Schema

The Agent should create a JSON input file from the conversation. The user does not need to see or edit this schema unless they ask for automation.